1. Field
This description generally relates to the field of speech recognition.
2. Background
Speech recognition systems in mobile devices allow users to communicate and provide commands to the mobile device with minimal usage of input controls such as, for example, keypads, buttons, and dials. Some speech recognition tasks can be a complex process for mobile devices, requiring an extensive analysis of speech signals and search of word and language statistical models. This is because mobile devices typically have limited computational, memory, and battery resources. As such, more complex speech recognition tasks are oftentimes offloaded to speech recognition systems located externally to the mobile device such as, for example, speech recognition systems in network servers.
Since more complex speech recognition tasks are performed on network servers and not on the mobile device, the results of the voice command may be limited to data stored in the network server. For these speech recognition tasks, the mobile device user does not have the benefit of viewing query results that may correspond to the voice command based on data stored in the mobile device. In addition, the delay time in transferring the voice command to the network server, performing the speech recognition operation at the network server, and transferring the query result from the network server to the mobile device can be significant. Significant delay time in the execution of applications on mobile devices, such as speech recognition tasks, can lead to a poor user experience.
Methods and systems are needed for performing speech recognition tasks on a client device, such as a mobile device, to overcome the above-noted limitations of speech recognition systems in mobile applications.